Especially as more and more high-powered AI-based computers and devices become more mainstream, this problem of efficient cooling is only going to become more pronounced. University of Missouri’s researcher, Chanwoo Park, has recently introduced what he calls the development of a novel cooling technology that would drastically cut on energy requirements.
“Cooling and chip manufacturing are two very related processes,” said Park, a professor at the Mizzou College of Engineering in mechanical and aerospace engineering fields. Some components fail when not cooled appropriately Tempers affect the efficiency of data centers Energy efficient data centers will define tomorrow’s artificial intelligence computing.
The cooling system dissipates heat from server chips through phase change, such as boiling a liquid into vapor in a thin, porous layer. Credit: University of Missouri
Innovating for the Future
Servers for data storage and processing are stored and processed in large data center complexes required for website and mobile applications, as well as cloud data. Still, these centers also are significant consumers of energy using more than 4 % of all electricity in the U. S. by 2022, where 40 % of this energy will be utilized for cooling equipment. This means that as the demand in the data center continues to be experienced, the energy consumption in the data center is also bound to rise.
To counter this, the U. S. Department of Energy has provided over $40 million to researchers that are working on the creation of new cooling solutions for data centers. Park recently acquired about $900 million. Slated to be funded from this initiative is COOLERCHIPS, which aims to get 65 million.
Today direct dispose of heat from computer racks in data centers either by air-moving fan or by liquids. Park and his team are working on the two phase cooling systems which involves the process of change of phase like boiling of a liquid and vaporization to cool the server chips. This system can work in a dormant state, drawing no power when less cooling is necessary and a little amount of power even when a pump is necessary.
“The liquid flow is random and on a thin metallic layer it evaporates,” said park. Thus utilizing this boiling surface, very reduced THERMAL RESISTANCE is got and efficient heat exchanging rate is done..
The system also has a mechanical pump that starts to pull more heat only when needed as part of the system. Preliminary conclusions show that the use of two-phase cooling procedures substantially decreases power consumption for refrigeration purposes.
The team is now engaged in designing and manufacturing of the cooling system with provision to connect and disconnect from the server racks. Currently, Park would like these systems to be in uses in the next ten years because computers are likely to be AI operated soon.
The current cooling systems will one day reach their maximum capacity, Park said. ‘They say we should prepare for the future and have a plan at least, and that is what we are attempting to do also, for the future of AI computing.’ This is a futuristic cooling system.
Park’s research relates to the goals of a new building on the campus known as the Center for Energy Innovation, which was established with an aim of allowing inter-disciplinary scholars to address energy issues arising from the surge in energy concerns and the fast growing artificial intelligence sector. The center is expected to major on the application of modern technologies in the generation, storage and usage of energy.
“The center will help us look for more concepts and development in this relation to efficient procedures,” Park added. It was quite an interesting experience to encounter such difficulties in resolving matters since these are rather multifaceted issues that can be addressed with the help of various professionals in the field only.
: I am eager to participate in the future collaborations.